Prehospital Fluid Administration for Suspected Sepsis in a Large EMS System: Opportunities to Improve Goal Fluid Delivery

Abstract Objectives Despite EMS-implemented screening and treatment protocols for suspected sepsis patients, prehospital fluid therapy is variable. We sought to describe prehospital fluid administration in suspected sepsis patients, including demographic and clinical factors associated with fluid outcomes. Methods A retrospective cohort of adult patients from a large, county-wide EMS system from January 2018-February 2020 was identified. Patient care reports for suspected sepsis were included, as identified by EMS clinician impression of sepsis, or keywords “sepsis” or “septic” in the narrative. Outcomes were the proportions of suspected sepsis patients for whom intravenous (IV) therapy was attempted and those who received ≥500 mL IV fluid if IV access was successful. Associations between patient demographics and clinical factors with fluid outcomes were estimated with multivariable logistic regression adjusting for transport interval. Results Of 4,082 suspected sepsis patients identified, the mean patient age was 72.5 (SD 16.2) years, 50.6% were female, and 23.8% were Black. Median (interquartile range [IQR]) transport interval was 16.5 (10.9–23.2) minutes. Of identified patients, 1,920 (47.0%) had IV fluid therapy attempted, and IV access was successful in 1,872 (45.9%). Of those with IV access, 1,061 (56.7%) received ≥500mL of fluid from EMS. In adjusted analyses, female (versus male) sex (odds ratio [OR] 0.79, 95% confidence interval [CI] 0.69–0.90), Black (versus White) race (OR 0.57, 95% CI 0.49–0.68), and end stage renal disease (OR 0.51, 95% CI 0.32–0.82) were negatively associated with attempted IV therapy. Systolic blood pressure (SBP) <90 mmHg (OR 3.89, 95% CI 3.25–4.65) and respiratory rate >20 (OR 1.90, 95% CI 1.61–2.23) were positively associated with attempted IV therapy. Female sex (OR 0.72, 95% CI 0.59–0.88) and congestive heart failure (CHF) (OR 0.55, 95% CI 0.40–0.75) were negatively associated with receiving goal fluid volume while SBP <90 mmHg (OR 2.30, 95% CI 1.83–2.88) and abnormal temperature (>100.4 F or <96 F) (OR 1.41, 95% CI 1.16–1.73) were positively associated. Conclusions Fewer than half of EMS sepsis patients had IV therapy attempted, and of those, approximately half met fluid volume goal, especially when hypotensive and no CHF. Further studies are needed on improving EMS sepsis training and prehospital fluid delivery.


Introduction
Sepsis is life-threatening organ dysfunction from a dysregulated response to infection (1). There have been high rates of sepsis-related mortality for decades, with 6.7% of deaths in the United States from 2005 to 2018 associated with a diagnosis of sepsis (2). In 2017, sepsis was the most expensive clinical condition treated in United States hospitals at $38.2 billion or 8.8% of total hospital expenditures, almost twice the cost of acute myocardial infarctions and stroke combined (3).
Early recognition and treatment of sepsis is necessary to prevent disease progression to severe sepsis or septic shock, which has an associated mortality rate of 28-54% (4)(5)(6)(7)(8)(9). A key component of early sepsis management highlighted by The Severe Sepsis and Septic Shock Management Bundle (SEP-1) is early intravenous (IV) fluid administration with a goal of at least 30 mL/kg intravenous crystalloid for patients in septic shock (10). Adherence to these guidelines is associated with better outcomes; however, this goal is met in less than half of severe sepsis and septic shock patients (11,12).
Emergency medical services (EMS) clinicians have a unique opportunity to accelerate the treatment timeline as they are often the first to care for sepsis patients, with 34%-75% of sepsis patients arriving to the emergency department by EMS (7-9, 13, 14). Previous studies examining prehospital intravenous fluid administration have shown an association between prehospital intravenous fluid administration and reduced mortality (15)(16)(17). Despite this opportunity to improve patient outcomes, prehospital intravenous fluid administration has only been reported in 23-64% of sepsis patients arriving by EMS (16)(17)(18)(19). Quantity of intravenous fluid administered is also widely variable, with data analysis from the 2014 Protocolized Care of Early Septic Shock (ProCESS) trial showing a median of 500 mL and interquartile range (IQR) of 225-1,000 mL at 31 centers across the United States (20), while the larger, nationwide, Dutch Prehospital Antibiotics against Sepsis (PHANTASi) Trial in 2018 found a smaller IQR (300-500 mL) with the same median fluid volume of 500 mL (19). There is a lack of evidence on the current state of prehospital fluid administration for sepsis and factors associated with optimal fluid delivery.
In this retrospective study, we investigated prehospital fluid administration in suspected sepsis patients in a single large EMS system. We also identified patient demographic and clinical factors associated with attempting IV fluid therapy and meeting EMS system fluid delivery goals prior to hospital arrival.

Ethical Approval
The institutional review board of the University of North Carolina at Chapel Hill approved this study by expedited review for research involving no more than minimal risk (IRB number: 21-0212; approved 5/3/2021). The IRB determined sufficient justification to waive the requirement for informed consent because this retrospective study involved no more than minimal risk and obtaining prospective consent was not possible.

Study Design and Setting
We conducted a retrospective cohort study of adult suspected sepsis patients transported by Wake County EMS, from January 2018 to February 2020. This time period was selected for two primary reasons. First, there were no major nationwide guideline updates or county EMS protocol changes to prehospital sepsis care within this timeframe. Second, this study aimed to examine typical prehospital sepsis care prior to the COVID-19 pandemic. The Wake County EMS system is a large, county-wide EMS system that responded to approximately 110,000 calls per year during the study period while serving a population of 1.1 million people. Patient care reports are created and maintained in a prehospital electronic health record (EHR) system (ESO Solutions, Austin, TX). The study data did not include patient names, dates of birth, or other direct identifiers. Encounter dates and times were removed from the final analytic dataset.
A suspected sepsis protocol was established at Wake County EMS in 2016 to direct prehospital screening and treatment of suspected sepsis patients (see online Appendix, supplementary materials). The adult sepsis protocol uses a sepsis screening tool that is a hybrid of variables from systemic inflammatory response syndrome (SIRS) and quick sequential organ failure assessment (qSOFA) criteria (21), as well as signs of poor perfusion. A patient screens positive if he or she has a suspected source of infection and at least one of the following: systolic blood pressure (SBP) <90 mmHg, heart rate (HR) >90/min, respiratory rate (RR) >20/min, Glasgow Coma Scale (GCS) score <15, or body temperature 100.4 F or <96 F. If a patient screens positive for sepsis, the treatment protocol recommends a 500 mL normal saline bolus that can be repeated up to a total of 2 L. The protocol allows the paramedic to administer fluids by IV or the intraosseous (IO) route. Fluid administration by IO in this patient population in this EMS system is rare. Therefore, this study focused on IV fluid administration. If the patient has two consecutive end-tidal CO 2 (EtCO 2 ) readings of < 26 mmHg at least 5 minutes apart at any point during the patient encounter, then a sepsis alert notification is given to the receiving facility prior to arrival.

Study Population
Eligible patients were 18 years of age or older who were treated and transported by EMS for suspected sepsis defined as paramedic primary or secondary impression of "sepsis/septicemia" or "septic shock" indicated from EHR dropdown menu, or keywords "sepsis" or "septic" in the patient care report narrative text. Study outcomes were the proportions of suspected sepsis patients on whom IV therapy was attempted, and of patients with established IV access who received 500 mL IV fluid (goal per protocol).

Study Variables
All variables used for analysis in the study were taken directly from the EHR data. Variables of interest included patient age, sex, race, ethnicity, comorbidities (congestive heart failure (CHF) or end stage renal disease (ESRD)), vital signs (SBP, HR, RR, GCS, temperature, and EtCO 2 ), and EMS transport interval (from departure from the scene to hospital arrival).

Statistical Analysis
The mean and standard deviation (SD) of patient age, and median and IQR of EMS transport interval were calculated. Categorical variables were described with percentages. Logistic regression models were used to estimate associations of outcomes (IV therapy attempted and fluid goal achieved) with demographics and clinical factors adjusting for transport interval. In multivariable logistic models, independent variables included age group (18-44, 45-64, 65-84, and >85 years), sex (male and female), race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, and non-Hispanic other), history of CHF, history of ESRD, abnormal vital signs (defined by the sepsis protocol described above), and transport interval (in 5-minute intervals up to >40 minutes). These covariates were considered a priori as potential predictors of the outcomes and were included in regression models. Adjusted odds ratios (ORs) and 95% confidence intervals (95% CIs) are reported. Confidence intervals not including the null value were considered statistically significant. However, given that this was a retrospective study that used all available data, we focused the interpretation of results on the magnitude of associations and their clinical importance. Statistical analyses were performed in SAS 9.4.
In adjusted analyses, female (versus male) sex, Black (versus White) race, other (versus White) race, and ESRD were negatively associated with attempted IV therapy (OR 0.83, 95% CI 0.72-0.95; OR 0.61, 95% CI 0.51-0.73; OR 0.55, 95% CI 0.31-0.97; and OR 0.58, 95% CI 0.36-0.95; respectively) ( Table 2). Having any abnormal vital sign during the EMS encounter was positively associated with having an IV attempt with EtCO 2 < 26 mmHg and SBP < 90 mmHg having the strongest associations (OR 2.41, 95% CI 2.06-2.81; and OR 3.33, 95% CI 2.77-4.02; respectively). There were also positive associations between longer transport intervals and IV attempts (Figure 2   95% CI 1.14-1.72; respectively). Transport intervals greater than 10 minutes were associated with increased rates of goal fluid volume administered ( Figure 3). Although there was a noticeable drop in the OR in the 35-40 minute interval, data were sparse at longer transport intervals, and this difference is likely due to imprecision of the estimate (i.e., wide CI).

Discussion
In over 4,000 suspected sepsis patients transported by EMS in a large EMS system, we found that fewer than half had IV attempts during their encounters with EMS, and approximately a quarter of all suspected sepsis patients received the minimum intravenous fluid bolus goal of 500 mL. Observed associations between fluid outcomes, patient demographics, and clinical factors suggest specific barriers to optimal prehospital fluid delivery. These findings suggest potential targets for intervention to improve fluid administration as part of the delivery of prehospital sepsis care.
In this study, we found that in sepsis patients, the IV attempt rate of 47% was higher than a previously reported rate of 37% in King County, Washington by Seymour et al. in 2012 (9). All abnormal vital signs evaluated in this study were associated with increased odds of IV attempts. This may indicate that EMS clinicians identified these patients as higher acuity, which is supported by Wang et al.'s findings, which indicate higher hospital admission rates for sepsis patients with abnormal vital signs in Boston, Massachusetts (7). Hypotension and low EtCO 2 were found to have the strongest association with an IV attempt, likely because these are the two abnormal vital signs that result in a more aggressive treatment pathway in the Wake County EMS suspected sepsis protocol.
While there is debate over the optimum volume of fluid administration in sepsis (22), the Wake County EMS suspected sepsis protocol goal of at least 500 mL of crystalloid solution is relatively conservative compared to Surviving Sepsis Campaign recommendations of a minimum of 30 mL/kg for septic shock (23), which is equal to 2,719 mL and 2,324 mL in the average US adult male and female, respectively (24). Despite these conservative fluid goals, we found only a quarter of suspected sepsis patients reached this target. Paramedics did appropriately recognize the relative importance of fluid resuscitation in hypotensive patients, as patients with SBP < 90 mmHg had the strongest association with reaching the target volume compared to other abnormal vital signs. Tachycardia and fever also had positive associations of reaching the fluid goal, though not nearly as strong. When intravenous fluids were given, we found that the median volume of 500 mL both met the protocol minimum treatment goal and matched the medians in previous prehospital studies reporting on the subject (18,19).
In our study, the average transport interval was approximately 16 minutes. Short transport intervals may be a contributing factor to the relatively low rate of suspected sepsis patients reaching the fluid goal in our study. The odds of an IV attempt seem to have increased with transport intervals greater than 25 minutes, which may indicate that time constraints contribute to the limited IV attempts during shorter transports. Similarly, we observed steadily increasing odds of reaching the fluid goal with transport intervals up to 35 minutes, though there were wide and overlapping confidence intervals in the longer transports. Though a positive association was identified between transport interval and both IV attempt and fluid goal success, our results indicate improvements in prehospital care are needed. Further studies should explore quality improvement initiatives addressing prehospital fluid administration, including clinician education and other implementation strategies. Given the positive trends observed with increasing transport intervals, EMS systems should consider transport interval as an important variable in compliance with quality improvement measures.
With published evidence of disparities in emergency care for women and racial minorities (25)(26)(27), it is important to reflect on what factors may have contributed to these groups being less likely to have IV attempts in this study. Akin to the recognition of implicit bias and systemic factors in differential emergency department care delivery (25), our findings suggest these are also important considerations for prehospital emergency care. Another potential contributing factor to observed sex and racial differences may be perceived difficulty of IV access in these populations. A large  ED-based study in 2020 by Shokoohi et al. found that both female and Black patients had higher rates of requiring ultrasound guidance to obtain IV access (28). Even after obtaining an IV, we found that female patients were less likely to receive their intravenous fluid volume goal. This finding raises a concern that female patients presenting with sepsis are being under-resuscitated compared to their male peers, especially when accounting for the previously discussed result that they are less likely to have IV attempts in the first place. While patient weight was not evaluated in this study, lower weight on average in female patients compared to males could lead to smaller overall fluid volumes while maintaining similar volume per kilogram overall.
The only other variable associated with significantly reduced odds of reaching the target fluid goal was in those patients with a history of congestive heart failure. While patients with this history may be more likely to be volume overloaded at baseline due to their underlying disease, sepsis causes a distributive shock that is more likely to result in a volume-responsive state. The available evidence shows mixed outcomes in CHF patients receiving at least 30 mL/kg bolus, but a trend toward improved mortality (29)(30)(31). With this available evidence, conservative fluid goals in this study, and a significantly lower proportion of CHF patients reaching their fluid volume goal, additional education for prehospital clinicians may be indicated to highlight the low risk of triggering an acute heart failure exacerbation with the relatively small volume of IV fluids being administered in the prehospital setting.
A 2022 study by Jensen et al. in Denmark found a lack of confidence among prehospital clinicians in deciding on the quantity of fluid to administer in suspected sepsis (32). They also found that almost all paramedics surveyed either agreed or strongly agreed that they were interested in more education about fluid therapy in sepsis. With the evidence demonstrating a lack of confidence in these decisions, and wide variability in fluid resuscitation identified in this study, additional educational opportunities for prehospital clinicians could potentially reduce variation in fluid delivery to patients. Promoting training and techniques to accelerate fluid administration in systems with short transport intervals also has the potential to improve the odds of reaching the target fluid goal, and increase overall fluid volume, during this limited period of time.

Limitations
There are several study limitations that are worth noting. Sepsis was broadly defined using prespecified impression options in EHR dropdown menus and keyword search of the narrative text. Given these broad criteria, patients included in the analysis may not have clinically required fluids nor had contraindications to fluid therapy that were not captured in these data. Since we could not confirm fluid eligibility, we did not expect all study patients to reach the target fluid goal, but still found a relatively low compliance rate. While we did identify a variety of demographic and clinical factors associated with both IV attempts and fluid delivery of at least 500 mL, we were not able to consider and adjust for all potential risk factors in this observational study using EHR data, such as clinical signs of overhydration or vascular congestion and relevant medical history. We were also unable to assess any patient-oriented outcomes such as morbidity and mortality due to a lack of availability of these data in the EMS EHR. With data from a single EMS system, results may not be generalizable to other EMS systems and regions, especially those with longer transport intervals. Future studies should examine why specific patients do not receive prehospital fluids despite EMS impressions of suspected sepsis and protocol guidance for fluid delivery. There would also be benefit to exploring variables that can improve the rate of fluid delivery to reach target fluid goal during the patient's limited time in the prehospital environment.

Conclusion
In suspected sepsis patients treated by EMS, fewer than half received IV attempts, and of those, approximately half achieved the protocol fluid goal. Female patients were less likely to have IV attempts and reach their IV fluid goal while patients with hypotension, tachycardia, and fever were more likely to receive both, after accounting for transport intervals. Black patients and those with ESRD were less likely to have IV attempts while CHF patients were less likely to reach the fluid goal. With these observed differences, further research on enhanced EMS sepsis training and techniques for rapid fluid delivery could increase optimal prehospital fluid administration for sepsis.

Disclosure statement
No potential conflict of interest was reported by the author(s).